The development of automated driving has brought a series of technical challenges, one of which being the safety of the automated vehicle. In automated driving Level 4 or above, an emergency fallback must be available for the automated driving system to bring the ego-vehicle to a stable and stopped state. Moreover, it remains to be investigated whether the application of this fallback causes additional risks for the ego-vehicle. This work introduces the speed-overriding emergency trajectory (SOET) as the fallback approach, and researches into the situations where this approach can be safely implemented. The concept and profile of the SOET is proposed first. Regarding the planning and implementation of SOET, two different strategies are explained and compared. Next, technologies on the state-of-the-art emergency trajectories are reviewed as a foundation of this research. Afterwards, general requirements which enable the basic functioning of the SOET are analyzed. The main focus of this work is the identification of situations which might be unsafe for the SOET to be applied. To carry out the identification, the definition of the term “situation” is clarified based on existing studies, as well as other related terms. The operational design domain (ODD) is specified to regulate the operating conditions for the SOET. On this basis, possible situations where the application of SOET might cause additional collision risks are derived. Another prerequisite for the identification is the methodology. First, a criticality metric is introduced, which serves as the judgement standard on whether the investigated situation is safe or not for the application of the SOET. The procedure of identifying the safety of the current situation is then sketched and explained in detail. Based on all the aforementioned efforts, the derived situations are analyzed in detail when it is unsafe for the SOET to be applied. In the end, the approach of SOET and the identification method is critically assessed, and the potential of applying the SOET is evaluated both within and outside the specified ODD.